orc

"Zug zug."  ·  Let the orcs do the work.

The senior-developer workflow, encoded as a Claude Code plugin — plan → debug → verify → ship — with hard guardrails that keep it off main, out of your git attribution, and entirely on your own machine.

Built for the senior-developer day. Safe enough for the security review.

Is it safe?  ·  Install  ·  Commands  ·  FAQ  ·  Examples

80 skills  ·  30 commands  ·  14 agents  ·  9 hooks  ·  0 telemetry  ·  MIT

---

orc is a full-SDLC workflow plugin for Claude Code: 80 curated skills, 30 composite slash commands, 14 specialist subagents, and 9 hook scripts that quietly enforce discipline (no commits to main, no AI-attribution trailers, destructive git commands gated, a dependency pre-flight, core rules injected at every session start). One umbrella command — /orc:flow — drives the full feature lifecycle from "I want to do X" to "PR merged", with the orc-implementer agent writing the code slice-by-slice in between — and the lifecycle no longer stops at PR-open: /orc:ci, /orc:release, /orc:deps, and /orc:incident cover what happens after.

It exists for one reason: every time a senior developer sits down to work, they should already know how the next hour goes — write the plan, watch the test fail, fix the cause (not the symptom), verify with evidence, ship the PR. orc encodes that loop.

Is it safe to use?

Yes — and you can verify every claim in this repo in about five minutes. orc is a workflow layer, not a service: no backend, no telemetry, no analytics. Network activity is limited to GitHub (fetching releases) and any MCP servers you add. Every guardrail is a short, readable shell script that Claude Code runs as a real hook — not a suggestion the model can talk itself out of: protected-branch commits and destructive git downgraded to confirm prompts, AI-attribution trailers refused outright, dependencies pre-flighted, the iron rules injected at every session start.

The full claim-by-claim breakdown — guardrail table, supply-chain hardening, FAQ — is in docs/safety.md.

What it does

orc maps the senior IC / tech-lead / architect day to a small set of composite commands. Most work fits this loop:

flowchart LR
    plan["/orc:plan"] --> start["/orc:start"] --> impl["implement"] --> qa["/orc:qa"] --> ship["/orc:ship"] --> ci["/orc:ci"] --> cleanup["/orc:cleanup"]
    cleanup -.->|"interrupted? /orc:resume"| plan
    cleanup -.->|"need status? /orc:status"| plan

Loading

Or skip the per-phase invocations and use /orc:flow to drive the whole loop — a gate at every phase, autonomous implementation in between via orc-implementer, and an interaction_policy ladder (manual/guided/auto) that decides how often it asks.

Outside the loop, commands cover the rest of the day:

• Fix & review — /orc:debug (root-cause first), /orc:code-review, /orc:address, /orc:stack-pr, /orc:fan-out
• Author — /orc:prd · /orc:trd · /orc:adr · /orc:rfc · /orc:postmortem · /orc:scaffold
• Operate — /orc:env, /orc:incident, /orc:deps, /orc:release, /orc:ci
• Track — /orc:jira, /orc:jira-breakdown, /orc:evidence, /orc:setup, /orc:triage, /orc:wayfinder
• Sessions — /orc:resume, /orc:status, /orc:cleanup — multi-phase work checkpoints to .orc/ and survives interruption

The full command table, the 14 agents behind it, and the 80-skill library live in docs/commands.md. Prefer learning by scenario? docs/examples walks 18 real situations end-to-end.

Install

Three commands from zero to working:

curl -fsSL https://raw.githubusercontent.com/HigorAlves/orc/main/cli/install.sh | sh   # 1. the orc CLI
orc doctor --fix                                                                       # 2. runtime tools (git, jq, + recommended)
orc install                                                                            # 3. register marketplace + enable plugin

Then restart Claude Code (or /reload-plugins) and type /orc: — the command palette should populate. That's the whole setup.

Tip

Pin a version with orc install --ref orc--v0.22.0, install without the CLI (/plugin marketplace add HigorAlves/orc), or build from source — the step-by-step guide with verification checkpoints, the marketplace-only path, and the full requirements table are in docs/install.md.

Statusline

While orc is enabled, Claude Code's status bar shows the work and the machine at a glance — no configuration needed (the plugin ships the default; your own statusLine always wins):

orc flow 6/9 implement │ PROJ-142 │ slices 3/7 │ auto │ ⎇ feat/checkout-retry* │ #128 ✔ approved
Opus 4.5 high think │ ctx ▓▓▓▓▓▓▓░░░ 68% │ $4.83 +412/−88 │ 5h 71%

Line 1 is the work: live orc session, Jira ticket, slice progress, autopilot level, branch, and the open PR with its review state. Line 2 is the machine: model + effort, an honest context bar, session cost, and your worst rate limit past 50%. It adapts to terminal width, honors NO_COLOR, and never breaks a render — and it feeds a context-monitor bridge that nudges the agent to checkpoint when context runs red. Tuning knobs in docs/configuration.md.

Iron rules (enforced by hooks + the using-orc skill)

1. No commits to main/master/develop — the PreToolUse hook downgrades them to a confirm prompt; approve only with explicit consent.
2. No code without a failing test first.
3. No claims without verification (run the command, read the output).
4. No fixes without a found root cause.
5. No AI attribution in code, commits, or PRs.
6. No multi-phase work without .orc/ checkpoints.
7. No silent broadcast in workspace mode — repo-touching commands need an explicit target flag or a confirming prompt.
8. No PR over the size budget (default 300 LOC) without a recorded choice — stack it, record a Size-budget-override: trailer, or abort.

Designed to stay lean

Claude Code loads every skill/command/agent description at session start (that's how it routes you), but loads a skill's body only when invoked. orc is built around that split: a ~3.5 KB always-on core (iron rules + routing), progressive disclosure for large reference skills, ≤200-char descriptions everywhere. Net: a fresh orc session pays only a few thousand baseline tokens before you type anything.

Docs

Page	What's in it
docs/install.md	Step-by-step install with verification checkpoints, marketplace-only path, requirements table
docs/commands.md	Full command table, the specialist agents, the skill library, the CLI reference
docs/examples	18 scenario walk-throughs — start here for usage
docs/configuration.md	Plugin settings, env vars, statusline tuning
docs/safety.md	The safety case in depth + FAQ
docs/architecture.md	Why the layout is what it is; hooks, state, and the .orc/ lifecycle
docs/contributing.md	Adding skills, commands, agents, and hooks

Development

See docs/contributing.md for conventions and docs/architecture.md for the why behind the layout. Developing on this repo? Load the plugin straight from your checkout instead of the marketplace:

claude --plugin-dir /path/to/your/clone/of/orc

Reload after edits without restarting: /reload-plugins.

License

MIT — see LICENSE.

_{Zug zug. Let the orcs do the work.}